Advanced Indexing and Search for Efficient Information Discovery

Promoting the sharing of technical data between researchers is one of the most effective ways to advance technology development. As a result, there is a growing trend for research communities to develop websites with curated catalogs of data sets and tools. However, once a large number of resources have been collected, it can become challenging to maintain a centralized site so that potential users can easily find resources that are useful to them.

We propose to address this problem by extending InferLink's ActiveSearch technology. ActiveSearch is a novel semantic search framework that combines natural language processing (NLP) techniques and a massive ontology so that users can search through document repositories in an intuitive fashion. In this project, we will extend this core technology to support search through collections of resources, such as data sets, software tools and analytics. This "catalog search" problem typically differs from document-centric search in some respects. For instance, there may be relatively little meta information associated with a resource. In addition, the search system must support expert researchers and contributors who are intimately familiar with the site's resources, as well as less sophisticated users who are unfamiliar with the resources.

Our primary application focus will be the IMPACT portal (Impactcybertrust.org) but the work will be generally applicable to other resource information sites. Due to the growing number of such sites, there are numerous commercial opportunities for a semantic search solution that can be easily customized for an application domain.